Mailing-List: contact cygwin-developers-help AT sourceware DOT cygnus DOT com; run by ezmlm
Sender: cygwin-developers-owner AT sourceware DOT cygnus DOT com
List-Unsubscribe: <mailto:cygwin-developers-unsubscribe-archive-cygwin-developers=delorie DOT com AT sourceware DOT cygnus DOT com>
List-Archive: <http://sourceware.cygnus.com/ml/cygwin-developers/>
List-Help: <mailto:cygwin-developers-help AT sourceware DOT cygnus DOT com>,
	<http://sourceware.cygnus.com/ml/#faqs>
Delivered-To: mailing list cygwin-developers AT sourceware DOT cygnus DOT com
Date: Thu, 5 Aug 1999 12:06:03 -0400
From: Chris Faylor <cgf AT cygnus DOT com>
To: "Larry Hall (RFK Partners, Inc)" <lhall AT rfk DOT com>
Cc: Corinna Vinschen <corinna AT vinschen DOT de>,
        cygdev <cygwin-developers AT sourceware DOT cygnus DOT com>
Subject: Re: ntsec: patch 9
Message-ID: <19990805120603.B1187@cygnus.com>
References: <3 DOT 0 DOT 5 DOT 32 DOT 19990805114647 DOT 00d3fbd0 AT pop DOT ma DOT ultranet DOT com> <37A9682E DOT EA185B11 AT vinschen DOT de> <37A8114F DOT 9101F2AE AT vinschen DOT de> <19990804214745 DOT A15316 AT cygnus DOT com> <37A9682E DOT EA185B11 AT vinschen DOT de> <19990805113216 DOT A973 AT cygnus DOT com> <3 DOT 0 DOT 5 DOT 32 DOT 19990805114647 DOT 00d3fbd0 AT pop DOT ma DOT ultranet DOT com> <19990805115304 DOT A1083 AT cygnus DOT com> <3 DOT 0 DOT 5 DOT 32 DOT 19990805120142 DOT 00d4c1d0 AT pop DOT ma DOT ultranet DOT com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.95.6i
In-Reply-To: <3.0.5.32.19990805120142.00d4c1d0@pop.ma.ultranet.com>; from Larry Hall (RFK Partners, Inc) on Thu, Aug 05, 1999 at 12:01:42PM -0400

Except that what Corinna describes is completely deterministic.

cgf

On Thu, Aug 05, 1999 at 12:01:42PM -0400, Larry Hall (RFK Partners, Inc) wrote:
>Usually during configures/builds though its intermittent and non-deterministic
>as far as I can tell.  Its pretty much the same situation Corinna describes.
>
>At 11:53 AM 8/5/99 -0400, Chris Faylor wrote:
>>When did you see the errors?
>>
>>cgf
>>
>>On Thu, Aug 05, 1999 at 11:46:47AM -0400, Larry Hall (RFK Partners, Inc) wrote:
>>>I see the same complaints from sig_send with the Aug 3 snapshot.  I use
>>>bash as sh too.  My CYGWIN variable only has nobinmode and ntea in it along
>>>with stuff to control the window title (I'm not at my machine now so I 
>>>don't recall the specifics of these settings).
>>>
>>>Larry Hall                              lhall AT rfk DOT com
>>>RFK Partners, Inc.                      http://www.rfk.com
>>>118 Washington Street                   (508) 893-9779 - RFK Office
>>>Holliston, MA 01746                     (508) 893-9889 - FAX
>>>                                        (508) 560-1285 - cell phone
>>>
>>>
>>>At 11:32 AM 8/5/99 -0400, Chris Faylor wrote:
>>>>I am not seeing anything like this.  I've just done a couple of configure/make
>>>>cycles with no problems.
>>>>
>>>>Do you have anything special in your CYGWIN environment variable?
>>>>
>>>>-chris
>>>>
>>>>On Thu, Aug 05, 1999 at 12:32:14PM +0200, Corinna Vinschen wrote:
>>>>>Chris Faylor wrote:
>>>>>> 
>>>>>> Thanks.  Applied.
>>>>>> 
>>>>>> Does the new snapshot still fail for you when you issue the
>>>>>> 'man tcsh' command?
>>>>>> 
>>>>>> cgf
>>>>>
>>>>>Hi Chris,
>>>>>
>>>>>unfortunately the answer is `yes'. I have found, that this behaviour
>>>>>is not reproducable beyond winsup-990726!
>>>>>
>>>>>Notice, that this happens regardless of the ntsec setting.
>>>>>
>>>>>winsup-990726 itself shows the behaviour:
>>>>>
>>>>>	tcsh> man tcsh
>>>>>
>>>>>shows man page, then pressing `q' in `less' results in:
>>>>>
>>>>>	0 0 [main] D:\bin\sh.exe 1029 sig_send: error sending
>>>>>	signal(-3) to pid 1029, Win32 error 6
>>>>>
>>>>>Error 6 is `illegal handle'.
>>>>>
>>>>>Since winsup-990801 it's worse than before:
>>>>>
>>>>>	tcsh> man tcsh
>>>>>
>>>>>... results in:
>>>>>
>>>>>	/usr/local/bin/groff: can't find `DESC' file
>>>>>	/usr/local/bin/groff:fatal error: invalid device `ascii'
>>>>>
>>>>>... and after pressing `q':
>>>>>
>>>>>	0 0 [main] D:\bin\sh.exe 1029 sig_send: error sending
>>>>>	signal(-3) to pid 1029, Win32 error 6
>>>>>
>>>>>If I try to run it with strace, I get the following on stderr:
>>>>>
>>>>>	strace.exe: couldn't get message length from subprocess,
>>>>>	windows error 6
>>>>>
>>>>>If, for example, the complete winsup directory is up to date,
>>>>>starting `make' results in:
>>>>>
>>>>>	make[1]: Entering directory `/src/cdkb21/winsup'
>>>>>	Making all in regexp...
>>>>>	make[2]: Entering directory `/src/cdkb21/winsup/regexp'
>>>>>	make[2]: Nothing to be done for `all'.
>>>>>	make[2]: Leaving directory `/src/cdkb21/winsup/regexp'
>>>>>-->	 0 0 [main] D:\bin\sh.exe 1009 sig_send: error sending
>>>>>	signal(-3) to pid 1009, Win32 error 6
>>>>>	make[1]: Leaving directory `/src/cdkb21/winsup'
>>>>>	make[1]: Entering directory `/src/cdkb21/winsup'
>>>>>	Making all in mingw...
>>>>>-->	 0 0 [main] D:\bin\sh.exe 1016 sig_send: error sending
>>>>>	signal(-3) to pid 1016, Win32 error 6
>>>>>	make[2]: Entering directory `/src/cdkb21/winsup/mingw'
>>>>>	make[2]: Nothing to be done for `all'.
>>>>>	make[2]: Leaving directory `/src/cdkb21/winsup/mingw'
>>>>>	Making all in utils...
>>>>>	make[2]: Entering directory `/src/cdkb21/winsup/utils'
>>>>>	make[2]: Nothing to be done for `all'.
>>>>>	make[2]: Leaving directory `/src/cdkb21/winsup/utils'
>>>>>-->	 0 0 [main] D:\bin\sh.exe 1020 sig_send: error sending
>>>>>	signal(-3) to pid 1020, Win32 error 6
>>>>>	make[1]: Leaving directory `/src/cdkb21/winsup'
>>>>>
>>>>>Let's talk about what happens: It's in EVERY case /bin/sh, that
>>>>>fails! In my environment, /bin/sh is bash. Regardless of the
>>>>>circumstances, it's only bash, that produces this error.
>>>>>If you look into the message, you will see, that it fails to
>>>>>work on a handle that references the calling process itself.
>>>>>
>>>>>I have attached the strace output of the above `make' example. It was
>>>>>compiled with -DDEBUGGING. I fear, it's not very useful because as
>>>>>ever when I try to strace the phenomenon, I get:
>>>>>
>>>>>	strace.exe: couldn't get message length from subprocess,
>>>>>	windows error 6